-
Notifications
You must be signed in to change notification settings - Fork 2.8k
Description
Which Umbraco version are you using? (Please write the exact version, example: 10.1.0)
10.4.0
Bug summary
We've observed that on one of our Umbraco 10 instances there's abnormal outgoing bandwith usage even with little to none incoming.
We believe this is cause by a infinite connection to /umbraco/previewhub
Specifics
What are we running:
- Standard Umbraco CMS
- Hosted on Azure using the Azure VM images (Windows Server 2022)
- Running in IIS 10, Not managed, Integrated Pipeline
- One site
- No loadbalancer
- 4 Cores, 16 GB RAM
- S2 Azure SQL
Comparing environments, one thing stand out, which is a long running call to /umbraco/previewhub as seen on the image bellow.

The suspicion that this is the cause is backed by our usage metrics which clearly shows that with each IIS Recycle the usage drops.
See the image below, that shows our scheduled recycles each night at 3 AM over a period of 7 days.

We've gone through the version comparer:
https://our.umbraco.com/download/releases/compare?from=10.4.0&to=10.5.1
Along with the info about V10.5.0:
https://our.umbraco.com/download/releases/1050
However, we can't find any note about issues with the PreviewHub, which is why we have yet to upgrade.
As mentioned previously, we have only observed this issue on 1 production instance, test and our other solutions does not have this issue.
The website in IIS has been configured with relatively default settings, only changing the URL Segment.

Upgrading Umbraco is not out of the picture but is a rather long process involving multiple parties.
So feedback or ideas beside upgrading is appreciated.
Steps to reproduce
We were not able to reproduce this behaviour on other solution.
Expected result / actual result
Expected result:
All connections should be terminated gracefully within a reasonable timeframe.
Actual result:
Connections live forever or until the App Pool is Recycled.